Skip to main content
. Author manuscript; available in PMC: 2024 Jun 1.
Published in final edited form as: Am Heart J. 2023 Mar 7;260:124–140. doi: 10.1016/j.ahj.2023.02.015

Figure 1.

Figure 1

Training and validation of causal machine learning model. Workflow of the repeated training and validation procedure for hybrid causal machine model consisting of the CT, TMLE, and net benefit modules. Each matched head-to-head OAC comparison data are randomly divided into 2 parts: 80% used for developing the clustering and estimating ATE and 20% used to validate the clusters. ATE, average treatment effect; CT, causal tree; OAC, oral anticoagulants; TMLE, targeted maximum likelihood estimation.